Dataset · camel-ai
Complex cross-domain tasks for reasoning development
camel-ai/loong
A small high-quality dataset of advanced mathematics and other disciplines for training models to reason.
What's inside
The English dataset contains between 1,000 and 10,000 text records for question-answering tasks. It was created as part of Project Loong and split into different subject areas, including advanced mathematics. The collection was gathered as a high-quality base for experiments on self-improving reasoning skills in language models.
What you can do with it
- Training and bootstrapping reasoning models on high-quality examples
- Evaluating LLM capabilities in solving complex cross-domain problems
- Tasks
- Question answering
- Languages
- English
- Size
- 1K–10K rows
- Downloads this month
- 1.8K
- Likes
- 64
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("camel-ai/loong")